de.lmu.ifi.dbs.elki.data

Class SparseByteVector

java.lang.Object

de.lmu.ifi.dbs.elki.data.AbstractNumberVector

de.lmu.ifi.dbs.elki.data.SparseByteVector

All Implemented Interfaces:

FeatureVector<Number>, NumberVector, SparseFeatureVector<Number>, SparseNumberVector, SpatialComparable

public class SparseByteVector
extends AbstractNumberVector
implements SparseNumberVector

Sparse vector type, using byte[] for storing the values, and int[] for storing the indexes, approximately 5 bytes per non-zero value (limited to -128..+127).

Author:

Arthur Zimek

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
static class	SparseByteVector.Factory Factory class.
static class	SparseByteVector.VariableSerializer Serialization class using VarInt encodings.

Field Summary

Fields

Modifier and Type	Field and Description
private int	dimensionality The dimensionality of this feature vector.
static SparseByteVector.Factory	FACTORY Static instance.
private int[]	indexes Indexes of values.
private byte[]	values Stored values.
static ByteBufferSerializer<SparseByteVector>	VARIABLE_SERIALIZER Serializer using varint encoding.

Fields inherited from class de.lmu.ifi.dbs.elki.data.AbstractNumberVector

ATTRIBUTE_SEPARATOR

Constructor Summary

Constructors

Constructor and Description
SparseByteVector(byte[] values) Create a SparseByteVector consisting of double values according to the specified mapping of indices and values.
SparseByteVector(int[] indexes, byte[] values, int dimensionality) Direct constructor.
SparseByteVector(TIntDoubleMap values, int dimensionality) Create a SparseByteVector consisting of double values according to the specified mapping of indices and values.

Method Summary

Methods

Modifier and Type	Method and Description
byte	byteValue(int dimension) Returns the value in the specified dimension as byte.
double	doubleValue(int dimension) Deprecated.
Vector	getColumnVector() Returns a Vector representing in one column and getDimensionality() rows the values of this NumberVector of V.
int	getDimensionality() The dimensionality of the vector space where of this FeatureVector of V is an element.
private int	getMaxDim() Get the maximum dimensionality.
Byte	getValue(int dimension) Deprecated.
private double[]	getValues() Returns an array consisting of the values of this feature vector.
int	iter() Iterator over non-zero features only, ascendingly.
int	iterAdvance(int iter) Advance the iterator to the next position.
byte	iterByteValue(int iter) Get the value of the iterators' current dimension.
int	iterDim(int iter) Get the dimension an iterator points to.
double	iterDoubleValue(int iter) Get the value of the iterators' current dimension.
float	iterFloatValue(int iter) Get the value of the iterators' current dimension.
int	iterIntValue(int iter) Get the value of the iterators' current dimension.
long	iterLongValue(int iter) Get the value of the iterators' current dimension.
short	iterShortValue(int iter) Get the value of the iterators' current dimension.
boolean	iterValid(int iter) Test the iterator position for validity.
long	longValue(int dimension) Deprecated.
void	setDimensionality(int dimensionality) Sets the dimensionality to the new value.
String	toString() Create a String representation of this SparseByteVector as suitable for SparseNumberVectorLabelParser .

Methods inherited from class de.lmu.ifi.dbs.elki.data.AbstractNumberVector

floatValue, getMax, getMin, intValue, shortValue

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Methods inherited from interface de.lmu.ifi.dbs.elki.data.SparseNumberVector

floatValue, intValue, shortValue

Methods inherited from interface de.lmu.ifi.dbs.elki.data.spatial.SpatialComparable

getMax, getMin

Field Detail

FACTORY

public static final SparseByteVector.Factory FACTORY

Static instance.

VARIABLE_SERIALIZER

public static final ByteBufferSerializer<SparseByteVector> VARIABLE_SERIALIZER

Serializer using varint encoding.

indexes

private final int[] indexes

Indexes of values.

values

private final byte[] values

Stored values.

dimensionality

private int dimensionality

The dimensionality of this feature vector.

Constructor Detail

SparseByteVector

public SparseByteVector(int[] indexes,
                byte[] values,
                int dimensionality)

Direct constructor.

Parameters:

indexes - Indexes Must be sorted!

values - Associated value.

dimensionality - "true" dimensionality

SparseByteVector

public SparseByteVector(TIntDoubleMap values,
                int dimensionality)
                 throws IllegalArgumentException

Create a SparseByteVector consisting of double values according to the specified mapping of indices and values.

Parameters:

values - the values to be set as values of the real vector

dimensionality - the dimensionality of this feature vector

Throws:

IllegalArgumentException - if the given dimensionality is too small to cover the given values (i.e., the maximum index of any value not zero is bigger than the given dimensionality)

SparseByteVector

public SparseByteVector(byte[] values)
                 throws IllegalArgumentException

Create a SparseByteVector consisting of double values according to the specified mapping of indices and values.

Parameters:

values - the values to be set as values of the real vector

Throws:

IllegalArgumentException - if the given dimensionality is too small to cover the given values (i.e., the maximum index of any value not zero is bigger than the given dimensionality)

Method Detail

getMaxDim

private int getMaxDim()

Get the maximum dimensionality.

Returns:

the maximum dimensionality seen

getDimensionality

public int getDimensionality()

Description copied from interface: FeatureVector

The dimensionality of the vector space where of this FeatureVector of V is an element.

Specified by:

getDimensionality in interface FeatureVector<Number>

Specified by:

getDimensionality in interface SpatialComparable

Returns:

the number of dimensions of this FeatureVector of V

setDimensionality

public void setDimensionality(int dimensionality)
                       throws IllegalArgumentException

Sets the dimensionality to the new value.

Specified by:

setDimensionality in interface SparseNumberVector

Parameters:

dimensionality - the new dimensionality

Throws:

IllegalArgumentException - if the given dimensionality is too small to cover the given values (i.e., the maximum index of any value not zero is bigger than the given dimensionality)

getValue

@Deprecated
public Byte getValue(int dimension)

Deprecated.

Description copied from interface: FeatureVector

Returns the value in the specified dimension.

Specified by:

getValue in interface FeatureVector<Number>

Specified by:

getValue in interface NumberVector

Parameters:

dimension - the desired dimension, where 0 ≤ dimension ≤ this.getDimensionality()-1

Returns:

the value in the specified dimension

doubleValue

@Deprecated
public double doubleValue(int dimension)

Deprecated.

Description copied from interface: NumberVector

Returns the value in the specified dimension as double. Note: this might seem redundant with respect to getValue(dim).doubleValue(), but usually this is much more efficient due to boxing/unboxing cost.

Specified by:

doubleValue in interface NumberVector

Specified by:

doubleValue in interface SparseNumberVector

Parameters:

dimension - the desired dimension, where 0 ≤ dimension < this.getDimensionality()

Returns:

the value in the specified dimension

longValue

@Deprecated
public long longValue(int dimension)

Deprecated.

Description copied from interface: NumberVector

Returns the value in the specified dimension as long. Note: this might seem redundant with respect to getValue(dim).longValue(), but usually this is much more efficient due to boxing/unboxing cost.

Specified by:

longValue in interface NumberVector

Specified by:

longValue in interface SparseNumberVector

Parameters:

dimension - the desired dimension, where 0 ≤ dimension < this.getDimensionality()

Returns:

the value in the specified dimension

byteValue

public byte byteValue(int dimension)

Description copied from interface: NumberVector

Returns the value in the specified dimension as byte. Note: this might seem redundant with respect to getValue(dim).byteValue(), but usually this is much more efficient due to boxing/unboxing cost.

Specified by:

byteValue in interface NumberVector

Specified by:

byteValue in interface SparseNumberVector

Overrides:

byteValue in class AbstractNumberVector

Parameters:

dimension - the desired dimension, where 0 ≤ dimension < this.getDimensionality()

Returns:

the value in the specified dimension

getColumnVector

public Vector getColumnVector()

Description copied from interface: NumberVector

Returns a Vector representing in one column and getDimensionality() rows the values of this NumberVector of V.

Specified by:

getColumnVector in interface NumberVector

Returns:

a Matrix representing in one column and getDimensionality() rows the values of this NumberVector of V

toString

public String toString()

Create a String representation of this SparseByteVector as suitable for SparseNumberVectorLabelParser . The returned String is a single line with entries separated by AbstractNumberVector.ATTRIBUTE_SEPARATOR. The first entry gives the number of values actually not zero. Following entries are pairs of Byte and Byte where the Byte gives the index of the dimensionality and the Byte gives the corresponding value.

Example: a vector (0,1.2,1.3,0)^T would result in the String
2 2 1.2 3 1.3

Specified by:

toString in interface FeatureVector<Number>

Overrides:

toString in class Object

Returns:

a String representation of this SparseByteVector

getValues

private double[] getValues()

Returns an array consisting of the values of this feature vector.

Returns:

an array consisting of the values of this feature vector

iter

public int iter()

Description copied from interface: SparseNumberVector

Iterator over non-zero features only, ascendingly. Note: depending on the underlying implementation, this may or may not be the dimension. Use SparseFeatureVector.iterDim(int) to get the actual dimension. In fact, usually this will be the ith non-zero value, assuming an array representation. Think of this number as an iterator. For efficiency, it has a primitive type! Intended usage:

 for (int iter = v.iter(); v.iterValid(iter); iter = v.iterAdvance(iter)) {
   final int dim = v.iterDim(iter);
   final double val = v.iterDoubleValue(iter);
   // Do something.
 }
 
 

Specified by:

iter in interface SparseFeatureVector<Number>

Specified by:

iter in interface SparseNumberVector

Returns:

Identifier for the first non-zero dimension, not necessarily the dimension!

iterDim

public int iterDim(int iter)

Description copied from interface: SparseFeatureVector

Get the dimension an iterator points to.

Specified by:

iterDim in interface SparseFeatureVector<Number>

Parameters:

iter - Iterator position

Returns:

Dimension the iterator refers to

iterAdvance

public int iterAdvance(int iter)

Description copied from interface: SparseFeatureVector

Advance the iterator to the next position.

Specified by:

iterAdvance in interface SparseFeatureVector<Number>

Parameters:

iter - Previous iterator position

Returns:

Next iterator position

iterValid

public boolean iterValid(int iter)

Description copied from interface: SparseFeatureVector

Test the iterator position for validity.

Specified by:

iterValid in interface SparseFeatureVector<Number>

Parameters:

iter - Iterator position

Returns:

true when it refers to a valid position.

iterDoubleValue

public double iterDoubleValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterDoubleValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position

iterFloatValue

public float iterFloatValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterFloatValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position

iterIntValue

public int iterIntValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterIntValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position

iterShortValue

public short iterShortValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterShortValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position

iterLongValue

public long iterLongValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterLongValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position

iterByteValue

public byte iterByteValue(int iter)

Description copied from interface: SparseNumberVector

Get the value of the iterators' current dimension.

Specified by:

iterByteValue in interface SparseNumberVector

Parameters:

iter - Iterator

Returns:

Value at the current position